"I must beg a more than ordinary degree of attention from those who will condescend to follow me through a narrow, rugged, uninteresting path, requiring a patience and labour which few will deem well bestowed on so vulgar a subject as an ALPHABETICAL CATALOGUE."

Mr. Panizzi to the Right Hon. the Earl of Ellesmere, 1848.

"In our precomputer systems much library work consists of mindless drudgery which is an insult to the intelligence of those who carry it out."

Michael Gorman / Cataloging and the new technologies, 1979.

The compilation of a library catalogue is, in all likelihood, an activity that is seldom performed for the amusement it brings. As Plato wrote in the Republic (Bloom, 1991, p.46) “Our need, as it seems, will make it.”1 A catalogue became a necessity.

Before the printed book, catalogues were descriptions of holdings not limited to manuscripts. For example, from a 1956 paper by Ruth French Strout: “A list from Exeter Cathedral, bearing the date 1327, records the opening words of each volume and gives its price. This document is part of a general inventory…” (pp.260-1). This suggests that other examples prior to this were similarly intended and were, therefore, not catalogues in the modern sense. Elaine Svenonius makes clear in her introduction to another paper, this time by Julia Pettee, written in 1936, that inventory lists were the first stage of catalogue development (Carpenter & Svenonius, 1985, p.73).2

The 16th century

A prime example of a catalogue in the early 1500s, chosen by Dorothy May Norris (1939, pp.126-30) and repeated by Strout (p.262), is that of Syon Monastery, Isleworth, extant in a manuscript of 1526. Typical of the time, this is a classed catalogue (one in which entries are grouped by subject) and the entries themselves being the “opening words of the second folio.”3 Norris also remarks that it is rather inconsistently done, but one innovation makes it remarkable: it is supplemented by an alphabetical author index. This seems such a useful, if not vital, addition that the only explanation of why it was not done earlier is that it was not worth the effort.4 There were not so many volumes of a given class in any library that a reader could not browse that subsection of the catalogue to find the author he (or she)5, required.

Certainly in this period, the best work in bibliography was not carried out by those who tended libraries. Natural philosophers and book dealers had a greater interest in taking some control over the works their occupations made use of. Swiss naturalist Konrad Gessner was perhaps the greatest example of the former; Andrew Maunsell, the latter. Gessner published an author bibliography (Bibliotheca Universalis) in 1545, complementing it with a subject index (Pandectarum) in 1548. Authors were listed by forename but Gessner also included a mechanism to search by surname. Moreover, he included cross-references of the see- format (“vide”), such as for alternate name spellings.6 Norris goes so far as to credit him with advancing the cause of cooperative cataloguing by his call for libraries to use his bibliography and index as their catalogues, simply adding shelfmarks as appropriate. An innovation that took hold much sooner was Gessner’s concept of providing more than one means to discover a book. With this principle in mind, in 1560, one Florian Trefler, who “believed no library was of any use without a catalogue,” took Gessner’s subject index as one element of his scheme for a five-part catalogue (Ibid., p.135).

Maunsell was an Englishman in the book trade. 35 years after Gessner he was sorting authors directly by surname and even stipulating the rules he followed for entries. Describing bibliography as a “tiresome businesse,” he was also motivated by necessity – in this case the need for English booksellers to have a catalogue “as the apothecary his Dispensatorium, or the schoole Master his Dictionarie” (Ibid., p.137). The Frankfurt book fair had its own catalogues, which soon developed into a proto-dictionary catalogue: a single alphabetic list comprising facets of author, subject, title; plus subject entries for places and events, and cross-references. Maunsell even implemented abstract subject headings in his bibliography (ie, ones not taken directly from titles). This is also presented as a single alphabetic list, making it even more the dictionary catalogue, long before it became conventional in the 1900s (Jackson, 1974, p.190).

These steps towards order must be seen in the wider context of a 16th-century that saw the destruction and scattering of books. Maunsell detected the need for what would later be called a code but, in Strout’s opinion, “the great scientists and philosophers of the age were not much aided in their endeavours by librarians” (1956, p.264). Tellingly, both Gessner and Maunsell only managed to write two of the three bibliographic works they had each originally intended to produce.

Nevertheless, for Krajewski, Gessner was “… The Father of the modern bibliography.” He lists many details (what might now be called metadata) and “undertakes an appraisal of the holdings.” The Bibliotheca Universalis was, unusually, author-alphabetical. The second volume – the Pandectarum – is a keyword list, ordered thematically, effectively becoming a scheme of knowledge or information organisation. The third, unrealised, volume was to be an alphabetical index of the Pandectarum. However, Krajewski’s interest in Gessner’s achievements stems more from his methods than his published works. He was the first to describe the use of paper slips for generating alphabetical lists in 1548. It is worth mentioning here that the first card catalogue, at Vienna, was not implemented until 1780. The primary use for sortable paper slips in Gessner’s day was excerption – a system allowing scholars to manage their notes and references. He specifies an algorithm for the processing of excerpts involving sorting paper slips, which are arranged as needed for a particular task. “This procedure describes a hybrid card catalogue in book form,” and a method of writing that itself contributes to the book flood, which that former technological solution will come to organise. To prepare his or her text an author need only rearrange notes (excerpts) to suit the subject at hand, copy them, then replace them in their original order for use another time, in another configuration.7

Krajewski’s thesis rests on the processing power afforded by sortable units of information. He makes a direct connection between paper slips and 0s and 1s; with moveable type, catalogue cards and punch cards between them in the sequence of related technological developments. Not only is the card catalogue a manual precursor of the computer, it is a machine in its own right – “information … available on separate, uniform, and mobile carriers … can be further arranged and processed according to strict systems of order” (2011, pp.3-14).

The 17th century

By the end of the 1600s Leibniz and others had started to put Gessner’s ideas into practice, but when considering the library catalogue at the beginning of the 17th century one must look to the work of Thomas James for Sir Thomas Bodley, at Oxford.

The foundation of the Bodleian

In 1598, retired diplomat and Hebrew scholar Bodley had written to the vice-chancellor offering to renovate and reconstitute the University library, which had been left so bereft after the events of the English Reformation that, by 1556, even the lecterns had been sold off. Approval was given, but Bodley was rarely in Oxford during the intervening years before the new library was ready to open on 8th November, 1602. In his place was James, a Fellow of New College who had been appointed by Bodley as Keeper in 1599, recommended by his on-going project to catalogue the Oxford and Cambridge college libraries. He was soon to be busy cataloguing Bodley’s acquisitions for the new library, a task that, inevitably, came to be more difficult than anticipated (Clapinson, 2015, pp.1-9).

James’s first printed catalogue was ready by 1605.8 It contained only 36 books in English amongst 5,611 main entries, and many more ‘analytical’ ones for works bound with others. “On this point Bodley was very vehement, and yet James … seems to have attached little importance to them and frequently failed to carry out Bodley’s wish”(Strout, p.265).9 Bodley was keen to keep standards high and, according to the academic convention of the day, was only interested in Classical texts in their original languages. Chaucer acceeded, but any available quartos of Shakespeare’s poems or plays would surely have been considered undesirable “riffe raffes.” These would have to wait for James’s successor John Rouse, and Robert Burton’s bequest of his library in 1640, for their accession to the shelves (Ibid., pp.14/31).

Bodley’s standards also applied to the cataloguing method to be adopted. James’s first manuscript catalogue was completed in 1601 before the library opened for use, but the 1605 printed catalogue was widely circulated and “presumably exerted influence” (Jackson, p.160). Bodley’s Statutes dictate that the Keeper is to “arrange the holdings according to faculties, preparing corresponding catalogs including full imprint….” The printed catalogue included call numbers and was “fundamentally a shelf list,” meaning one page was dedicated to each shelf. Shelf position was determined by classification under the modern scheme of the four faculties: Theology, Medicine, Jurisprudence, Arts. (Norris, p.144; Jackson, pp.159-60).

Given this correspondence between shelf contents and catalogue page, the individual pages were indeed used as shelf lists on the ends of the presses.10 At this early stage the shelves had spare capacity; and so did the pages. The empty space was ready for accessions but was also used for related entries; for example, where to look for similar works. Call numbers included shelf position and so needed amending if enough new books were added. Jackson lists earlier experiments in compiling call numbers without shelf or even bookcase locations but none was truly “relative-locational” and it is not known if Bodley was aware of them. Some anonymous works were entered under title, which was treated like a subject heading, with similar titles (for example, by place name) listed together. The typically long titles of the age were often abbreviated in this process, but place and date of publication were normally included. Quarto and octavo formats were designated in the call number, as they were stored separately, but folios had no extra reference. An index to the catalogue listed authors and titles in an alphabetical list (Jackson, pp.161-2; Norris, pp.144-5).

Norris and Jackson both describe various inadequacies and inconsistencies in the 1605 catalogue. For example, one “deficiency” was its reliance on title terms, which could vary greatly for works on a single subject, meaning that not everything covering the same topic would be found in one place. James retired after publishing a revised and expanded catalogue in 1620 but continued compiling subject guides to individual faculties for the use of undergraduates. Seemingly looking to future publications, there were many headings with few entries; out of 29 subdivisions under Geometry, one contained only one entry (with the heading therefore identical to the title). It would later became a rule for dictionary and card catalogues that this should not occur, after the concept that subject headings should have a degree of generality (Jackson, p.163).

Soon after the Bodleian was established, German preacher Georg Draud was publishing bibliographies with novel characteristics, including measures to improve subject access. For example, including out-of-print works; see-references after outdated terms to current equivalents; and “a striking prototype of the corporate entry.” Draud implemented catchwords – a word from a title to ‘catch’ the work for a particular class.11 Germany in general gave primacy to the author-alphabetical catalogue (Stichwort), with a subject index (Schlagwort). ‘Term’ catalogues (those of booksellers) had differing priorities to bibliographies for libraries. Firstly, they were only concerned with what was currently in print; secondly, language was important, more so than to libraries. Format information is also of more interest to those keeping books than selling them on. Jackson even compares Draud’s “atomistic” subject approach to postwar coordinate indexing using ‘uniterms’, which worked on early computers working with data on punch cards (Ibid., pp.193; also see Norman, 2012).

In the preface to the 1620 Bodleian catalogue James recommended leaving classification to the catalogue and shelving the books in simple alphabetical order by title. While this might not seem progressive to readers today, used to well-established classification schemes ordering books on library shelves for the benefit of browsing, it should be remembered that, until the public library boom in the 19th century, stacks remained closed to readers.12 James’s alphabetical method was the one preferred by library staff who frequently had to find and replace specific titles quickly. German scholar J. H. Hottinger observed that “when shelves are closed to the public it does not matter very much what the (classification) scheme is” (Ibid., p.181), though it could be argued that it was not his job to fetch the books.13 Some Continental libraries took organisation by shelving to extreme lengths in the later 17th century, but the larger ones were more conservative and Le Tonnelier of Paris declared in 1677 (agreeing with the James of 1620) that “the great variety of content as well as format made exact subject control via shelving impracticable and that the catalog was the only answer” (Ibid., p.197).

Advice on establishing a library

Parisian pamphleteer-cum-librarian Gabriel Naudé sought to influence library policy in his 1627 work Avis pour Dresser une Bibliothèque, which was popularised in translation to English by John Evelyn. He recommends both a systematic subject and alphabetical author catalogue (‘divided’ in the same volume); evidence that these were still more often ideas than realities. He also advised that shelving arrangements must allow for expansion (Strout, p.265), though Jackson comments that he is less concerned with methods of organising books than with acquiring them for his patrons (chiefly Cardinal Mazarin), seemingly unaware of the “expansion just over the horizon,” though the Mazarine library came to be the “largest library which had ever been brought together in the world” (p.170). “When Naudé enlarged the collections at the Mazarine in the 1640s the acquisitions so far outran the cataloguing that access actually did depend – though not by plan! – upon a systematic arrangement…” (Ibid., pp.196-7).14 Nevertheless, Strout credits Naudé with the modern idea of catalogues “as a means of finding books and identifying them bibliographically” (p.265).

Scotsman John Durie undertook his “drudgery” in 1649 in England for the King’s Library, though its patron would very soon be without his head. He recommended a subject catalogue, subdivided by language, and printed, with regular supplements; no fixed locations; and no need to clutter the catalogue with less valuable books, though no need for them to be “cast away” either. He also advised that university “doctors” should assess which new works should be acquired and catalogued, something that Humphrey Wanley, of the Bodleian, was to repeat as the 17th century drew to its close. By this point the Bodleian’s 1620 and 1674 author catalogues were exceptions to the tendency to organise by subject, particularly on the Continent, perhaps excepting Germany. However, all such were now as much finding aid as inventory, and the importance of uniformity was being realised, with Thomas Hyde’s 1674 Bodleian catalogue introduced by the first explicit code of rules.15 Printed catalogues and supplements were the norm, yet by 1850 Panizzi would be declaring them impossible (Norris, p.179). His solution would owe a lot to the work that Leibniz was doing in the mid-to-late 1600s, after Gessner and Draud.

Krajewski has Leibniz and his colleague at Vienna, Hugo Blotius, putting Gessner into practice “to provide efficient catalogues.” Leibniz had acquired an excerpt cabinet and used it to order notes he had made to further his scholarship and professional life. ‘“Thus he also maintained his library according to topical order, without regard for different formats.”’ (Von Murr, 1779, quoted in Krajewski, p.19). Leibniz’s catalogue for Wolfenbüttel (1691-9) involved “an assistant cutting, sorting, and gluing … a directory that was to remain, into the 20th century, the only general author catalog of the Duke August Library.” Leibniz had reported on the need for a “logical or scientific catalog” to satisfy readers who ask for works, not by author, but by subject, demonstrating that a librarian – even Leibniz himself – by this point can no longer be expected to know the collection intimately (Ibid., pp.16-21).

The important lesson from these cases is that, whether the subject or author catalogue is considered the primary one, it has been largely accepted that, where a catalogue exists, the one must be supplemented by the other, at least in index form. Subject access has become more sophisticated; in contemporary terms, there are multiple access points. A radically effective and efficient way of facilitating this was tantalisingly close to be being realised.

The 18th century

Though the 1700s saw great advances in general, catalogues did not advance much until it was drawing to a close and were, again, inadequate to the demands of the burgeoning and newly systematised sciences, technologies and industries. “It may have been because libraries were so busy with their growing collections that they now ceased to philosophise about what catalogues ought to be.” However, towards the end of the century in post-Revolutionary France (1791), theoretical advances were made by the new government, despite its intention to be purely practical. A national code was written requiring the creation of a card for each confiscated book, with the author’s surname underlined for the purposes of filing. If there was no author, then a keyword in the title was to be chosen. This, with the various other requirements such as number of volumes, statement of illustrations, etc., which essentially remain in current codes, makes this code a “steppingstone to the extensive cataloguing developments of the next century” (Strout, pp.266-7). Norris adds: “Not only have we instructions here for a card catalogue, but also for accessioning and shelf guiding” (p.195).

For Krajewski, “… at the beginning of the 18th century it is by no means self-evident that a library should own a directory of its holdings. To find a book on a certain subject, one usually follows the classified shelving of books.” Leibniz, however, was arguing that this situation was untenable; “a library without a catalog … resembles the warehouse of a businessman who cannot keep stock” (1679). Krajewski extends the metaphor to one of lost profit due to lack of information. Leibniz’s “plan anticipates registering every book merely once, precisely on a slip of paper … Theoretically, this procedure could have made numerous catalogs with the same data set.” In practice, only the author catalogue was completed due to lack of resources (2013, p.22), and though paper slips were used they were always to be rebound in book-form. However, the potential is there for (further) recombination. The slip represents the book and can, in many cases, be referred to instead of the book itself (Ibid., p.23).

The foundation of the British Museum

This institution was founded by Act of Parliament in 1753, bringing together the collections of Hans Sloane, including some 40,000 printed books and 7,000 manuscripts, with two other library collections – those of Sir Robert Cotton and the Earls of Oxford. Four years later in 1757 these were joined by the Royal Library – the collections that had been assembled by various British monarchs up to George II.

An anonymous Proposal for building a Royal Library, and establishing it by Act of Parliament had been published around the turn of the century, possibly by the Keeper of the Royal Library, Richard Bentley. “Urged first were a new building and residence for the keeper, the preferred location being praised for its elevations and “dry sandy ground” and fire-protective distance from the nearest structure” (Jackson, p.213). Jackson also notes that Cardinal Borromeo’s (Ambrosiana, Milan), James’s and Hottinger’s writings on running a library all include the advantages of buying collections in bulk. The Proposal recommended ‘“a perpetual yearly revenue for the purchase of books.”’

The British Museum library officially opened in 1759 at Montagu House. Pre-existing catalogues of the Harleian collection were on sale “but the reader had no comprehensive index to consult until an alphabetical author catalogue was printed in 1787 – the work of library officers in their two daily “vacant hours” (Norris, p.200). The holdings were arranged as the shelf space in the several rooms seemed to recommend; not until 1790 was a shelf list produced.” By 1774 a new reading room offered 120 seats, more light and less damp (Jackson, pp.214-6).

North America and Continental Europe

At this point some libraries in the New World had the right works in their collections to tell them where they stood with respect to libraries in Europe. For example, Harvard owned James’s 1605 Bodleian catalogue, Draud’s 1611 classified bibliography of Latin and German publications, Bale’s 16th-century English national bibliography; all of which did much to demonstrate the limitations of its own collection. Its 1790 catalogue listed 12,000-13,000 volumes; certainly the largest collection in the Colonies. This is after it recovered from a fire destroying everything that wasn’t on loan (more than 90%) in 1764. “The only work expected of a librarian at a colonial college which might be considered “professional” was the preparation of a catalog.” However, the presidents of both Yale (1743) and Princeton (1760) compiled their library’s catalogues. Such efforts inspired the donation of books, but their collections remained so small that “there was no urgency to adopt even the few refinements displayed in the Leyden catalog of 1595 and the Bodleian guides of 1605 and 1674.” Yale started assembling subject catalogues in 1740 but there was no awareness of advances at court libraries in Germany such as that of Leibniz (Ibid., pp.219-22).

At the Herzog August Bibliothek Leibniz served the Dukes of Brunswick-Lüneberg as librarian, amongst other offices, while pursuing his own projects of private scholarship. As a book user he was most interested in what was, naturally, useful. For librarianship this meant that his approach was one of creating the best, highly curated, current resource as opposed to collecting everything, or giving priority to classics. He scoured any available “bookseller and fair catalogs, correspondence, the Journal des Sçavans, and the Royal Society Transactions;” but also saw the value of a library being able to purchase collections on the market wholesale; something he complained about not having the resources to do. In 1689 in Italy, on a mission for his employers, he “composed a list of some 2,500 titles, arranged first by subject and then roughly by date; based on both classics and recent science.” It was a starting point rather than a comprehensive bibliography. His 1702 dossier to his patron, Rudolph Augustus, listed the “increasing rate of book production, the rising costs of books and binding16, the state of the library’s holdings, and comparable data from other collections.” He also suggested several positive ideas to raise money. It was under the succeeding Duke, Anton Ulrich, that plans for the new library building progressed. Book acquisition had increased, leading to the need for temporary storage. A vignette of a typical day in the new library is provided by Jackson (p.247): “Despite the librarian’s devotion to service to “all,” a traveller of 1731 – eight years after the books had been placed there – found just a few scholars using the collection, the staff absorbed in niceties of cataloguing.” In general, “… provision of appropriate space seldom kept up with the acquisitions.” Another contemporary diarist records “crowding and disorganization” typical of the 100 libraries he visited in Germany, the Low Countries and England between 1709 and 1711 (Ibid., pp.245-7).

The German term for shelf list is Standortskatalog, meaning ‘location catalogue’, but because these were separated by subject it became common in larger libraries to also keep an Accessionskatalog, which kept everything in one place. New documents such as “learned society transactions, academic dissertations and learned journals confronted the cataloger with numerous volumes of multiple authorship.” The solution was to prepare entries for parts of composite works and pamphlets. This had been proposed at the Bodleian (not for the first time) in 1697. But, “Such analytics were to become familiar and valued in many libraries… .” (Ibid., pp.258/261-2).

Moving towards mobile data

Krajewski notes the “gradual turn away from fixed shelving and toward more delocalized addressing” (pp.31-2). If, due to the rate of book acquisition, each book can no longer be allocated a specific address relating to the shelf on which it is to be found, then these ordering and addressing tasks are shared by the catalogue. The transition is difficult to imagine for readers or librarians not used to discovering books thanks solely to their systematic physical arrangement, plus an exposition of their bibliographical data in a shelf list. Even if that physical arrangement holds, sheer numbers can make finding a book impractical without first resorting to a catalogue, which, after all, is an abstract and thus more wieldy representation of the collection.17 The question is how one searches. Up to this point, whatever catalogues exist, the convention is that a systematic arrangement of books will literally guide the searcher to his (or, less likely, her) destination: the book in question or at least a book on the subject in question. Libraries are now reaching the point at which the searching should be done on paper first, which should significantly narrow the physical area to be searched in order to obtain a book. In its turn, the catalogue is under pressure to accommodate new titles in their respective places, breaking up the linear list and forcing mobility on the bibliographic references. This is rather revolutionary. The programme of the Enlightenment and the Encyclopædists was still to classify under the assumption that this procedure was a process of discovery – of understanding the world as it was – rather than the imposition of an arbitrary, man-made, order; something that loose paper slips or cards seemed too ephemeral or fluid to represent.

Yet at some point in the 1700s someone discovers that playing cards are a practical media for storing and sorting units of information. Krajewski shows that one Abbé François Rozier, while applying the techniques of Gessner in his task to tabulate the publications of the Académie des Sciences, make the small step from using slips of paper to ‘“cartes a jouer,”’ which were then produced blank on one side and, “beside their widespread availability, their uniform measurement allowed easy shuffling. They were also able to withstand “robust handling” (p.33).

The Josephinian catalogue18, for the Hofbibliothek in Vienna (now the Österreichische Nationalbibliothek), is “considered the first card catalog in library history.” Krajewski reports, qualifies, and then endorses this statement, making sure to stipulate three reasons why it should be distinguished from Leibniz and Rozier’s early modern paper slip techniques.19 These are: “written instructions for the cataloger, a division of labor organized around interfaces, and the duration of the catalog” (p.39). It is stored in some 205 boxes designed to look like bound volumes and comprises 300,000 cards. The boxes may have been a gesture to divert attention from the fact that there was to be a further stage – that of using the slips to create bound author-alphabetical and subject catalogues, estimated to take up some 50 to 60 folio volumes. But the task was too great to contemplate, and the interim card catalogue became the library catalogue. “Thus, a failed undertaking tacitly turns into a success story” (Ibid., p.42-3). This project was initiated in 1780.

Blotius’s legacy was felt throughout Viennese libraries, however. Franz Stephan Rautenstrauch’s catalogue at Vienna university library predates the Josephinian catalogue in its use of procedural instructions and paper slips by two years. Furthermore, this one manages to become a bound catalogue (the paper slips are also maintained).20 Also, its rules are more generalisable, resulting in them becoming “an imperial standard for catalogs, over thirteen years before the regulation generally cited as the first of its kind,… the efforts to create a French national bibliography in 1791” (Ibid., p.44).21

In this period the abolition of the Jesuit order had created a minor book flood for the Austrian libraries developing their paper slip techniques, but the confiscation of sacred and aristocratic collections in the French Revolution, though on a different scale, was to be managed in similar fashion, with the French innovation of using cards. However, the plan fails and, “In the end, not a single volume of the planned national bibliography is printed on the basis of playing cards.” Its success is that the principle of standardisation is established, even if the cards are still seen, despite the situation at the Hofbibliothek, as merely a step on the road to a bound catalogue (Ibid., p.47).

The 19th century

This period saw major advances in cataloguing, albeit somewhat later than the equally major advances in book production, which could now be said to be on an industrial scale.22 Nevertheless, there was plenty of time for philosophy. In the early 1800s significant disputes about the merits of classed catalogues took place, which would spend the rest of the century waiting for the Library of Congress to implement abstract subject headings, after Charles A. Cutter’s Rules For A Dictionary Catalog of 1876. One of the disputes concerned what should happen at the British Museum.

Before Panizzi started work there, in cataloguing, William Croswell had delivered a catalogue for Harvard on paper slips in 1817. What is most interesting about this is that, according to Krajewski, he devised the method out of necessity without any knowledge of Gessner or his early modern disciples, nor of the Josephinian catalogue. Tasked with compiling the Harvard catalogue single-handed, Croswell had spent nine years getting dizzy and making little progress. With a final deadline looming he hit upon the idea of cutting up the previous 1790 catalogue and interleaving its entries with manuscript ones for later accessions. It was left to his successor Joseph Green Cogswell to expound the true significance of Croswell’s achievement, noting that his system is extensible and reconfigurable – ‘“the work is done for ever; it may be increased so as to embrace all the books ever printed, without requiring any part of it now done, to be done again”’ (Krajewski, pp.74-9).

The catalogue of the British Museum library23

Consisting of four previously separate collections, with various and inconsistent catalogues, the British Museum’s Trustees recognised that a new and complete catalogue was necessary to render the collections useful to the public. Nothing achieved in the 18th century proved satisfactory and, once George III’s collection (The Royal Library, now known as the King’s Library) had been added in 1823, another attempt was made, this time by the Rev. T. Hartwell Horne, which also failed.24 The next proposal was to give the job to Antonio Panizzi, who had been appointed Extra-assistant Librarian in 1834. As part of a Royal Commission, a Select Committee started investigating the British Museum’s administration the same year, instigating a lengthy and well documented debate on author versus “classed” catalogues.

Panizzi argued against a classed catalogue – “The continual discoveries in science make classification ridiculous” – but for a subject index to the author catalogue.25 He also contended that learned society transactions and periodicals should not be entered under subject but new “fixed heads” and also under author. Entries should be transcripts of title pages. For consistency, one person should be “in charge.” Admittedly, an author’s name might need to be found, but that could be achieved “by using standard bibliographies.”

In 1837 Panizzi was appointed Keeper of the Printed Books yet debate continued about who would superintend the catalogue and how it should be done, leading Panizzi and his colleagues to draw up cataloguing codes that were unified into the ‘91 Rules’, which were adopted by the British Museum in 1841 and widely praised as a decisive advance.26 Carpenter (p.1) states that “All modern codes [that is, up to AACR2] descend from” these 91 Rules. Main entry “stands for the one principal entry for a book. References are made to this principal entry, and the rules explain the different classes of references.” To give an apposite example, Rule I reads:

I. Titles to be written on slips, uniform in size.

The 91 Rules are also noted for giving “great authority to the title page” (Strout, p.268). Subject entry always derives from the title page somehow. This is a long established practice, albeit codified more thoroughly than ever before. Carpenter states that the 91 Rules “continue practices found as far back as the catalogues for libraries in medieval monasteries” (p.2).

Work started in 1839 and the A-volume was printed in 1841. It was the last to be printed. Panizzi had only reluctantly agreed to printing volumes successively and the problems he had predicted arose. For example, “necessary cross-references were continually springing up after the volume had gone to the printers.” In 1847 another Royal Commission was appointed to investigate the cataloguing programme and its lack of progress. Panizzi testified that it would take until 1895 to produce a printed catalogue of the library’s state in 1854; yet changing the method would mean having to start again (they were up to the letter D). Since 1841, entries had been made in manuscript (ie, by hand) interleaved with the printed Ellis and Baber catalogue of 1819, though it was in poor alphabetical order. In 1849 library employees W. Croker and E. Roy suggested new entries to be written on paper slips, inserted in guard books. Panizzi adopted this and “set a staff on to copy all the titles. [and]… In 1851 the catalogue in 150 volumes was put into the Reading Room” (Norris, pp.208-12).

In the previous year Panizzi had taken the trouble to write to Francis Leveson-Gower, First Earl of Ellesmere, in his capacity as chairman of the Royal Commission.27 Panizzi’s task was to catalogue some 500,000 works for a manuscript catalogue that would then be reproduced in print.28 Amongst Panizzi’s points are that space is a concern, even for the catalogue – calculating the prodigious extra volume required by cards as compared to finer paper slips. He demonstrates that the shelves must be consulted; that earlier catalogues cannot be relied on; and that it is impossible to proceed letter-by-letter, leaving each one complete. He claims:

“There is no instance of a catalogue equal in extent to that of the British Museum collection of printed books having been attempted, and there is no precedent for rules being embodied to carry into execution even a compilation far inferior in extent.”

Crucially, he also shows that a printed catalogue of an increasing library can never be complete. For this reason, the working catalogue in the reading room ought to be in manuscript for it to best serve its purpose (Carpenter & Svenonius, pp.15-47).

Panizzi held on to his job and was even appointed Prinicpal Librarian in 1856. Neverthless, in 1861, the Trustees still wanted a printed catalogue, as Cambridge University had managed to achieve. However, at the British Museum the original 150 volumes of guard books was well on the way to becoming 3,000 (Norris, p.213). In 1842, when Panizzi was in the middle of his difficulties with superiors who expected a printed catalogue, Cogswell’s successor at Harvard, Charles Folsom, reverted that institution to a bound catalogue, but, “in doing so, he apparently learns a lesson” and introduced one Charles Coffin Jewett to the concept of the paper slip catalogue (Krajewski, p.79).

By this mid-point in the 19th century the issues being discussed are those of ‘“modern”’ cataloguing (Strout, p.270). She contends that, in 1850, cataloguing practice in the US can be said to advance beyond that in Europe with Jewett’s code for the catalogue at the Smithsonian. Jewett accepted Panizzi’s 91 Rules with some modifications relating to entry. For example, corporate bodies would be entered under their names without any of Panizzi’s intermediate ‘form’ headings such as ‘ACADEMIES’. Also, his method for dealing with anonymous works rejects the British Museum’s conventional approach of finding a catchword from the title to define the work by subject. In general, Jewett was prescriptive, yet efficient, producing only 33 regulations. Cutter’s Rules of 1876 supported Jewett in his differences with Panizzi. Cutter makes explicit that the catalogue’s goal of being able to show what a library has by any given author requires pseudonymous works to be entered uniformly, even if the title page does not contain the ‘authorised’ name of the author.

As Librarian and Assistant Secretary at the Smithsonian Institution, Jewett believed that its library was essential to fulfilling its stated objective of promoting the ‘“increase and diffusion of knowledge among men.”’ More then creating a great national library, however, Jewett was motivated by the possibility of creating a great national catalogue – “a union catalogue of the holdings of all public libraries in the Untied States.” In justifying this he invokes the spirit of Leibniz:

“How much this would promote the progress of knowledge… how much, by rebuking the rashness which rushes into authorship, ignorant of what others have written, and adding to the mass of books, without adding to the sum of knowledge.”

Like Leibniz he is also more interested in books relevant for progress rather than for their own sake, and Svenonius points out that his desire for a universal catalogue distinguishes him from Panizzi, with his large but essentially local project. Jewett proposes that the contemporary printing technology of stereotyping would allow bibliographic records to be stored in print-ready form to be reused whenever required.29 Jewett appreciated and argued for the level of standardisation that this would require, also because his vision was, by definition, one of cooperative cataloguing. What he was proposing was essentially discrete bibliographic records that could be sorted to suit either an author-alphabetical (his preference, supplemented by a subject index) or a classed catalogue. Like his illustrious predecessors, Jewett laments how poorly understood the task of compiling a catalogue is, describing what “may seem a light task,” as “arduous and perplexing.” Interestingly, in explaining the cataloguing code he advocates, Jewett notes that some “conform more to the rules advocated by Mr. Panizzi, than to those finally sanctioned by the Trustees of the Museum” (Carpenter & Svenonius, pp.48-61).

Cutter would argue that such a “legalistic” approach to cataloguing was mistaken; it is rather “an art that applies a few highly generalized rules by analogy to specific cases.” Cutter’s main concern was convenience to the user (Ibid., p.49). Jewett made further refinements to the British Museum’s 91 Rules, which Cutter “crystallized” in his Rules. These relate to corporate (ie, collective) authorship and mark a distinction between Anglo-American and European practice; Cutter’s Rules became the standard in the Anglo-American region but were not embraced in Germany (Jackson, p.380). Jackson also suggests that American libraries, being both not so large as the great European ones and more accessible to the general public (as opposed to the educated elite), naturally preferred author-alphabetical catalogues over systematic/classified ones. Some US libraries reported having a classified index to an author-alphabetical catalogue.30 Jewett gave less importance to the subject catalogue than Cutter, believing that most library users arrived with specific authors in mind. Pragmatically, (and like Leibniz) Cutter wanted the subject catalogue to stand in for (and assist) the library staff in finding suitable works for readers on any given subject.

At about the same time Billings (later to be of the NYPL, then at the Surgeon General’s Office) produced a card-based alphabetical subject catalogue by rearranging cards that had been prepared for the compilation of a printed author catalogue. These “masterpieces” were published in 1879 (Index Medicus) and 1880 (Index Catalog of the SGO, vol. 1). Now that the increasing rate of acquisitions was making the printed format’s perpetual incompleteness a serious problem, the card catalogue, common since the mid-1850s for current acquisitions, started to become the singular reference of a library’s holdings. Cambridge used cards from 1861; Harvard opens its to readers the same year; France and Germany followed (Ibid., pp.382-5; Krajewski, p.80).

The 20th century

The decision at Harvard to open the card catalogue to readers was made by library assistant Ezra Abbot, who at the time had his own assistant in Charles A. Cutter. By 1904 Cutter’s Rules for a Dictionary Catalog were in their fourth edition. Somewhat ironically, Strout cites the same year (in which took place an international meeting in St Louis) as the one in which it was acknowledged that card catalogues “had won out over the other forms” (p.273). The dictionary catalogue had been the common US format in the latter part of the 19th century and into the 20th (Jackson, p.406). A printed volume (or volumes), it presented authors, titles, series, and subjects interfiled in one alphabetical list. For the reader, this was certainly an advance over earlier classed catalogues; even ones supported by indexes (Norris, p.225). As a printed and bound catalogue, however, and especially for large and expanding libraries, it was subject to all of the difficulties so far illustrated; difficulties that the card catalogue was superseding.

Cutter’s status developed in large part due to his production of the highly regarded catalogue of the Boston Athenæum (1874-82). Treating cataloguing more as art than algorithm, he regretted that the availability of printed cards from the Library of Congress since 1901 – something he had done as much as anyone to make possible – marked, somehow, the end of cataloguing.31 His Rules were relevant yet, both because the Library of Congress had only recently begun distributing pre-printed cards and also because he had grappled the philosophical questions so well that his work would be a reference for cataloguing authorities in the future, however fewer in number they might be thanks to his codifying efforts.

However, his credo, “The convenience of the public is always to be set before the ease of the cataloger,” is not necessarily at odds with legalistic approaches. Panizzi and Jewett didn’t write all of those rules for their own convenience but for the success of the catalogue.32 And, as Svenonius presciently noted in 1985, online catalogues allow rigidly constructed bibliographical records to be presented to the searcher in flexible and convenient ways. Cutter’s “Objects” [objectives] have been “restated by [Seymour] Lubetzky and in the Paris Principles,” marking, as they do, a turn away from long lists of specific rules towards establishing fewer and broader principles, as a basis for cataloguing codes.33 Cutter’s Objects also repeat Leibniz’s statement about both author and subject indexes being necessary. Cutter is also credited with properly instantiating corporate authorship into cataloguing, contrary to the “German practice” of treating such works as anonymous (Carpenter & Svenonius, pp.62-7), and paving the way for Library of Congress Subject Headings (LCSH). After centuries of countless alternate systems of classes, filled by catchwords from titles in a specific collection, LCSH is a pre-coordinated, and largely enumerated, controlled vocabulary, alphabetical list of headings” (Frické, 2012, p.200).34 These allowed readers searching card catalogues to find books classified under a huge array of specific (and modular – hence “pre-coordinate”) headings). This could only have been achieved by allowing one authority – in this case the Library of Congress – to determine both the headings and how to apply them to the US bibliography, then distribute catalogue cards depicting these decisions to smaller libraries The next step was for Melvil Dewey, his Library Bureau and American Library Association, to standardise and commercialise the apparatus of card cataloguing. As Krajewski illustrates, index cards become the basis of information organisation not just in libraries but throughout the commercial world,35 until they are replaced by their natural successors, computers.

In 1985 Michael Carpenter wrote that, at the time of the ICCP in 1961, “there was some awareness that computers might change cataloguing” (Carpenter & Svenonius, p.177). A few years later, what came to be called automated cataloguing arrived. In the words of Harris, “The Library of Congress has developed a system of Machine-Readable Cataloging (MARC) by which complete catalog cards can be transmitted on electronic tapes and printed out by a receiving library” (pp.272-3). Electronic tape was state-of-the-art and many computers of the day were still using punched tape as data storage medium. More importantly, libraries, including the Library of Congress, were still using catalogue cards.36 Though MARC has lasted to this day – and is not at immediate risk of being ousted by BIBFRAME37, its putative linked open data successor – it was designed to produce printed cards; its particular innovation being a standard for encoding bibliographic data that a computer could manipulate.

In her 1996 book Future Libraries, Future Catalogues, Pat Oddy is critical of MARC’s limitations, claiming that it, “in its parody of the two-dimensional catalogue card, perpetuates the muddying of the distinction, quite clearly made in the cataloguing code itself [AACR2], between intellectual work and physical manifestation” (p.94). The concept of main entry, which was first codified in the 91 Rules, makes a lot of sense for a physical catalogue, even in card form, but has some undesired consequences, such as separating rather than collocating manifestations of a single work that have variants of their author’s name, though that be one person.38 Machine-readable bibliographic records, nevertheless, were an essential step in making possible what has come to be known as the online catalogue, which would by its very nature undermine the concept of main entry. However, it could be said that, due to MARC, the process of cataloguing was not fundamentally changed by computerisation, though perhaps that change is now beginning. It is more certain that the computer changed the catalogue itself.


This concludes the second chapter concerning the evolution of the library catalogue since 16th century. Next, the subject changes to computerisation and the library warehouse.



1. More classically rendered as the proverb necessity is the mother of invention, or simply, needs must. back

2. The second being finding lists; the third, bibliographic tools. Krajewski, Norris, and again, Svenonius write about the early catalogue being an inventory or simple list of holdings. back

3. Meaning second page. back

4. Strout attributes an earlier example to Johann Tritheim of Germany, though his was a bibliography rather than catalogue (p.262). back

5. There were also nuns at Syon but they apparently had a separate library, a catalogue for which does not survive. back

6. Which would have been much more common in his day, as evidenced by the variations in his own name (Gessner / Gesner). back

7. Incidentally, the software this author chose to manage his notes in this study (Scrivener) uses the same principle – separate documents (graphically represented by index cards) that can be sorted in order to produce a composite draft. back

8. The Bodleian is frequently credited as having the first printed catalogue. Leiden apparently had one (Nomenclator) ten years earlier in 1595 but it is not known whether it was general (ie, covering all classes). back

9. Jackson suggests that multiple works in one volume were the main reason James thought it unsatisfactory to shelve by subject, but he understood that they could be separated in a catalogue (p.164). back

10. It seems that Bodley decided on this after increasingly exasperated attempts to get James to catalogue adequately (see Norris, ch.7). In the end he demanded that James "listed each book as it stood on its shelf" (Norris’s words, p.143). back

11. Common in the 19th century – for example, stipulated by Panizzi – until Cutter made the case for abstract subject headings. back

12. Frické (p.12) observes: "Alphabetical arrangement helps searching but not browsing; but, in contrast, form or thematic arrangement helps browsing but not searching." back

13. The following chapter will show that warehousing methods allocate shelf positions dynamically or, in some cases, randomly. back

14. A systematic shelving scheme the University of Strasbourg in 1613 had been attempted in order to render a catalogue unnecessary. back

15. And a lengthy complaint about "the difficulty and weariness of the task" (Norris, p.150). She attributes the preface to Hyde and the catalogue itself to one Emmanuel Prichard. Julia Pettee (writing in 1936 and cited in Carpenter & Svenonius, pp.72-89) attributes modern advances to Hyde’s 1674 Bodleian catalogue, including using only one form of name for an author, including translations, and cross-referencing pseudonymous works to the authorised form. Cutter followed suit in his Rules, first published in 1876. back

16. Books were supplied unbound until about 1830 (p.205). back

17. Norris (pp.184-9) gives a representative example of the 17th-century catalogue in that of the Sion College library, 1724. For each entry, a capital letter denotes the press (and the subject), a Roman numeral the shelf, and an Arabic figure in the margin represents "the numbers of the books on the shelves." back

18. "named for Austria’s "enlightened despot" Joseph II" (Wright, 2014, p.33). back

19. Although the work was done on paper slips, which were then pasted onto cards (Krajewski, p.41). back

20. Johann Wilhelm Ridler, in 1823, renews the ’basic catalogue’, complaining of decline and even slips being used "to the foulest ends by library officials." It is maintained to this day for works published before 1931. back

21. For example, Norris credits the French Code of 1791 as the first national code (p.195). back

22. Jackson quotes statistics from a 1911 issue of the Bulletin de l’Institut International de Bibliographie. World book production: in 1838 – 27,838; in 1858 – 65,190; in 1887 – 100,000. Periodical production showed a rate of increase at least three times greater. back

23. See Norris, ch. 10. back

24. The same Horne produced a classed catalogue for Queen’s College, Cambridge in 1827 but Norris (p.219) demonstrates that his scheme is "confused and difficult to follow." back

25. Norris also argues that another reason against a classed catalogue is that it would show where the library’s collection was deficient (p.224). Oddy (1996, p.27), writing as Head of Cataloguing at the British Library, claims that "the lack of a coherent means of subject access to its collection is a problem which haunts the British Library to this day." back

26. Rules for the Compilation of the Catalogue, British Museum (1841); reproduced in Carpenter & Svenonius (1985). back

27. Carpenter claims that the Commission’s "task was supposed to be the impeachment of Panizzi for insubordination" (p.16). back

28. For comparison, in 1849 the number of printed volumes at the Bodleian was some 220,000, though many contained multiple works. back

29. In mid-19th century printing a stereotype was an embossed plate from which copies could be printed on a press. The plate was made from a mould of a forme of type, itself comprised of moveable type, plus leading, etc. The advantage was that the pages only needed to be made up once, then stereotyped, allowing the moveable type to be reused while copies could be printed indefinitely from the stereotyped plate. Jackson comments that, in practice, the clay mould often deteriorated before a satisfactory plate could be produced (p.385). back

30. In a systematic subject catalogue, works are listed alphabetically by author only under each subject, making the author inferior to the subject. back

31. Cutter even described cataloguing as an "innocent pleasure," an untypically positive description when compared to those from other noted practitioners (Carpenter & Svenonius, p.65). back

32. Cutter in fact states that "the convenience of the public" and "the ease of the cataloger. In most cases they coincide" (Carpenter & Svenonius, p.66). back

33. See Pettee in Carpenter & Svenonius, pp.72-89. The ’Paris Principles’ (formally Statement of Principles) are the outcome of the International Conference on Cataloguing Principles (ICCP), 1961. They formed the basis for the Anglo-American Cataloging Rules (AACR), and others. back

34. Svenonius points out that it is also faceted into Topic, Place, Time and Form (p.179); Oddy that LCSH "are the single most important example of a de facto standard that is not based on any codified set of principles." back

35. Interesting examples being the Carnegie Steel Company – later to fund many public library buildings – and the Tabulating Machine Co. (later IBM). back

36. The OCLC ceased production of catalogue cards on October 1, 2015 (OCLC, 2015). Production had peaked in 1985. back

37. Peters (1991, p.78) notes the inefficiency of storing duplicate MARC records all over the Western world, something that BIBFRAME is expressly intended to address using machine readable linked open data for bibliographic records on the semantic web. back

38. Cutter moved to prevent this by use of an authorised form of name, even if that meant deviating from the title page. For works of multiple authorship, including variations across multiple editions, useful collocation becomes more problematic. back